Shubhankar Pinge
Product Engineer
Founder, Ramen Reserve
I build AI tooling and drive adoption through integration — meeting people inside the systems they already work in. At Spectrum, I lead AI adoption for the product org behind the #1-rated telecom app.
Outside work: Dispatch, a multi-agent chat platform with intent-based routing and MCP integrations.
Projects
AI products and automation I've designed, built, and shipped.
View Full Portfolio
Dispatch
askdispatch.chatSmarter tech. Smaller footprint.
A production multi-agent assistant. A hybrid dispatcher answers directly, or hands off to one of eight specialists — each request routed to the right-sized model.
- Auditable routing
- Only high-confidence transfers; failures fail open. Scored in promptfoo against a 90-case golden set.
- Right-sized model per request
- Agents resolve to a model by specialty tag across OpenRouter, with cost tracked per call.
- Platform guardrails
- Invite-gated auth at the database layer, usage metering, quotas, and SSRF-guarded custom tools.
Sous Chef
Source-backed recipes that rescale on request.
Planner
Three days in Kyoto, from 35 grounded sources.
Adjust times, swap activities, or ask for transport and dining detail.
MoDecks CLI
modeckscli.comDecks that close. From the command line.
A concept landing page for a terminal-native deck generator, presented as a six-slide pitch inside a faux PowerPoint shell — sidebar thumbnails, a Present mode with fullscreen + keyboard nav, and a live typing terminal that “renders” a board deck. Zero-dependency vanilla HTML/JS; the animated Mo logo sting is hand-rolled SVG driven by an authored-time easing model.
- Deck-as-website
- Six 16:9 slides rendered from a fixed 960×540 canvas and scaled to any viewport; scroll-reveal, active-slide tracking, and a real Present mode.
- Hand-built motion
- Logo sting recreates a spiral→M morph with a stroked path, spinning O that shrinks into the full stop, and a rally-arrow accent — pure SVG math, no animation library.
- Mailing list on the edge
- Signup modal posts to a Cloudflare Worker backed by D1 (first/last/email, validation, dedupe), CORS-locked API, mobile-safe (no iOS zoom, no forced keyboard).
A Ramen Reserve Inc. project
Decks that close. From the command line.
One-line prompt in. Board-ready deck out. Circulated before EOD.
Furniture-shopping demo for ChatGPT: an MCP server exposing two read-only tools that recommend furniture for a room from a photo or description, rendered as verified in-chat product cards. Includes an evaluation harness covering the API contract, MCP protocol, and model behavior.
Technology company behind Dispatch and MoDecks CLI. Its site is four static pages, no framework, ~8 KB gzipped, with a pop-out mailing list running on a Cloudflare Worker and D1.
Resume platform with 15+ ATS-friendly templates, drag-and-drop section editing, live preview, and PDF export. Real-time sync via Convex, with anonymous access that migrates into a registered account — no signup wall.
Scrapes ratings, reviews, and rankings from the iOS and Google Play stores into PostgreSQL, exposed via a Flask interface and API — competitive intelligence on autopilot.
Interactive dashboard for App Store KPIs — rating insights, satisfaction trends, and competitive analysis in customizable views.
Puppeteer automation that logs into test accounts and captures full-page screenshots of key pages — consistent UI records for QA and documentation.
Professional Experience
Seven-plus years building internal tooling, AI systems, and the measurement behind them.
(Charter Communications)
Product Engineer | AI Tooling & Automation
Customer Commitment org — CX, SRE, Testing, and VoC
- Customer Commitment — joined the team consolidating CX, SRE, Testing, and VoC to build its measurement foundation. Defined and standardized the metrics, reporting, and processes now driving defect resolution across all four teams.
- JAWS — built end-to-end. A chat widget over whichever agentic CLI and local MCP servers a user already runs, model- and tool-agnostic, with org context pushed over the air. Terminal-level capability, approachable interface.
- MCP integrations — wired into the tools teams already use, paired with hands-on enablement and adoption metrics. The on-ramp to an AI-assisted development lifecycle, adopted as a transition rather than a mandate.
- Call center analysis — rebuilt from a 90-day queue into a 2-hour AI-assisted pipeline: pulls data, diagnoses the issue, ranks fixes by impact vs. effort, models call savings, generates the exec deck.
- Defect alerting — new tickets routed to owning teams within 15 minutes of filing, with live detection and resolution SLA dashboards leadership pulls from directly.
- Cox merger — adapted the measurement stack for the $34B Charter–Cox merger: VoC collection for rapid issue resolution, app rating tracking, and defect tracking extended to the incoming portfolio ahead of close.
(Charter Communications)
Product Engineer | Reporting & Analytics Lead
Self-Service Products & Platforms — 16 product programs
- App ratings — owned the pipeline, forecasting, competitive intelligence, and strategy. Held the #1 telecom app rating four years running, the org’s Market Leadership goal and a target carrying 15–20% of annual outcomes. 200+ apps tracked as a production data service.
- AI optimization (AIO) — researched how AI models perceive and rank consumer brands, and what it takes to influence them. SEO’s successor, now that assistants are where customers discover products.
- Built org-wide delivery tracking over 12,000+ Jira tickets; the org hit 140% of committed top-priority projects against a 90% goal.
- Standardized reporting across 16 programs, cutting weekly update time 75% and execution discrepancies 74%.
- Built screenshot automation for any test account — since adopted by Quality, Design, CX, and Product.
(Charter Communications)
Product Analyst | Reporting & Analytics
- Automated the monthly competitive app report across 72 iOS/Android listings — a 2–3 hour manual process down to under a minute.
- Built the company’s first app rating forecasting model, plus an iOS review scraper for data no third-party tool exposed. Adopted by other orgs.
- Migrated program reporting from manual Excel to automated Tableau, cutting KPI cycles from 4+ days to under 2.
- Built a Jira-connected Gantt template with instant refresh — days of chart building down to under one. Standard tooling from 2024 on.
Project Delivery Analyst
- Analyzed large datasets for customer experience insights that drove system and process improvements.
- Improved a Voice of Customer classification model and its data tagging for more reliable reporting.
- Built an offensive lineman ranking metric from NFL Next Gen data on AWS.
(Charter Communications)
Data Analyst (Voice of the Customer)
- Developed a new metric tracking how defects drag down Customer Satisfaction (CSAT).
- Implemented and evaluated NLP solutions for feedback classification and sentiment analysis.
Data Analytics Associate
- Supported data analytics and reporting initiatives.
Engineering Associate
- Contributed to data-driven engineering projects.
Skills & Expertise
A showcase of my technical capabilities and certifications.
Python
TypeScript
JavaScript
React
NextJS
Tailwind
Puppeteer
Workflow Automation & Webhooks (n8n)Web Scraping & Browser Automation
PostgreSQL
AWS
Snowflake
Databricks
Git
Firebase
Tableau
PowerBIAcademics
My educational background and research contributions.
Formal Education
B.S. Industrial Engineering
M.S. Information Technology
Research Projects
Designing and implementing an RFID Personnel Tracking System within the Penn State: Learning Factory.
Tackled inefficiencies in outpatient procedures by analyzing causes of high Patient Length of Stay (LOS), developing solutions, and presenting findings at the MCREU Annual Conference.
Co-authored paper "Investigating Customer Needs for an Information Artifact for Emergency Department Patients" (Proceedings of Industrial Engineering Research Conference, May 2016).
Let's Connect
I'm always open to discussing new projects, creative ideas, or opportunities.